Business Context
A hospital uses a stricter standard than the industry default for monitoring sterilization defects. Instead of the standard 3-sigma limits, they require 2-sigma limits to catch potential issues earlier. They also want to force the analysis even if subgroup sizes vary slightly or don't match the nominal plan perfectly.
Data Preparation
Simulating weekly sterilization batches. Counts are low, but we want to detect minor elevations that wouldn't trigger a standard 3-sigma alarm.
Copied!
| 1 | |
| 2 | DATA mycas.SterilityChecks; |
| 3 | INPUT WeekID BatchSize Defects; |
| 4 | DATALINES; |
| 5 | 1 100 0 2 100 1 3 100 0 4 98 1 5 100 0 6 100 3 7 100 4 8 100 1 9 100 0 10 102 1; |
| 6 | |
| 7 | RUN; |
| 8 |
Étapes de réalisation
1
Run cChart with sigmas=2 (tighter), no3SigmaCheck=True (allow tests on non-standard limits), and allN=True (include all batches).
Copied!
| 1 | |
| 2 | PROC CAS; |
| 3 | spc.cChart / TABLE={name='SterilityChecks'} processValue='Defects' subgroupValue='WeekID' sigmas=2.0 no3SigmaCheck=true allN=true primaryTests={test1=true}; |
| 4 | |
| 5 | RUN; |
| 6 |
Expected Result
The control limits should be narrower than standard (calculated at 2 sigmas). The data points at Week 6 (3 defects) or Week 7 (4 defects), which might be acceptable under 3-sigma, should likely be flagged as outliers (Test 1) due to the tighter `sigmas=2` setting. All weeks, including 4 and 10 (varying BatchSize), must be included in the analysis due to `allN=true`.